For most projects, both at work and in my sparetime, I’m using a mix of Markdown and Sphinx/ReStructuredText for creating READMEs, reference documentation, and topic guides. While this setup works for me, I’m not the biggest fan of some aspects of ReStructuredText and have therefore started looking more and more at asciidoc. The downside this is that, though, I hadn’t found any framework similar to Sphinx that allows to work with a collection of asciidoc-documents. Well, and then I stumbled upon Antora this week.
While Sphinx is focused on taking a specifc folder full of rst files and converting it to HTML, TeX, PDF, what-have-you, Antora is a more generic framework for mixing documentation stored in multiple locations and converting it to HTML. Out of the box it supports building documentation for multiple releases of a project based on tags, branches, and so on.
The output is based on a playbook defining what modules should make up the final documentation and how they should be converted into the desired output format. Each module can define its own navigation and in turn can contain multiple components with pages and assets like images. The modules referenced in a playbook don’t have to reside in the same code repository. For instance, there could be one team working on the user guide in their own repository while another team focuses on the developer guide in another.
Playbooks now instruct the converter what UI bundle (zip files with markup templates, stylesheets etc.) to use to generate HTML out of these modules and what navigation should be rendered on the final output.
Coming from Sphinx it definitely took me some time getting comfortable with this structure. In general, Antora by default expects a certain file system structure (detailed in Organizing Your Documentation for Antora and right now can only deal with “packaged” bundles. For the latter there already exists a ticket on Gitlab π
Antora in general looks despite its young age extremely useful and I’m pretty sure I will integrate it in one of my next projects. I already tried to convert one with only a couple of pages and really liked the resulting structure. Perhaps we finally have a real competitor to Sphinx in the market π
Do you want to give me feedback about this article in private? Please send it to comments@zerokspot.com.
Alternatively, this website also supports Webmentions. If you write a post on a blog that supports this technique, I should get notified about your link π